Migrating from mvBase to D3

Hi Guys,

So we're migrating to D3 we have had initial success following the online documentation but just wondering if anyone else out there has done the same and has any tip or advise for a smooth transition.

Thanks in advance

Steve

Hi Guys,

So we're migrating to D3 we have had initial success following the online documentation but just wondering if anyone else out there has done the same and has any tip or advise for a smooth transition.

Thanks in advance

Steve

Did you get any documents from @Claude Marinelli?

We have a policy of having one set of programs.  That is, the same source has to compile under D3 or Mvbase.  This means we've had to augment our source to look at a "flag" to determine if the machine is D3 or not.  This comes into play when opening a file:  OPEN 'SYSTEM,ACCOUNT,WORKFILE' for Mvbase or OPEN 'ACCOUNT,WORKFILE,' for D3.

Let's say you have 3 subroutines:  A, B, C.  Subroutine A has a COMMON statement as does subroutine C.  A  calls B is fine.  But, if B then calls C this will crash with bizarre results.  You need to include the COMMON in subroutine B.

I would submit that with our applications Mvbase performed much, much faster than D3.  As a simple example I have a file which is 16MB.  It takes several seconds under D3 to COUNT, SELECT or whatever - even when the same process is repeated.  Under Mvbase it was instantaneous.  It seems if the file is memory resident with Mvbase, processing is super fast.

D3 is more stable than Mvbase.  But, when something has gone awry you need Rocket support to resolve.  

Become familiar with role of the USERS file and the MDS file.  There is another file in D3 known as FSI:MDS,, This is sort of like a mirror image of the MDS file. When the D3 system reboots, the items from this file replace those in the MDS file. If you take the password out of the MDS item, you need to remove it from this other file, too.

I would recommend becoming familiar with the switch that controls the flavor of Pick on D3.  We didn't know about it at the time of implementation and so were tripping over things like the order of port, account and user in a "who" call.  

The content of a verb definition is quite different in D3 so programs which examine the MD for the content of an entry needs to be modified.  

Our applications evolved over time and had a lot of General Automation R91-esque.  D3 is much better at handling background jobs, for instance.

It takes a little bit of getting used to the differences of the TCL stacker.  Sometimes I like the D3 way and sometimes I don't.